Re: [Razor-users] TeS algorithm

[Rob <ca...@cs...>]

On Mon, Oct 28, 2002 at 01:28:46PM -0500, Jeff Bailey wrote:
> This feels like another case of security through obscurity though.  It would
> seem that having a public review of the algorithms would not only lead to
> fixing possible holes in the algorithm, but would also continue to lead to
> the development of more Free Software that can only help Razor.  We already
> know that we'll always have to adapt to new spamming techniques.  I don't
> think that hiding the TeS algorithm will slow that down at all.

I feel very much the same way.  I understand the point about not wanted
to allow spammers to abuse the system, but there has been a good deal
of research done into distributed trust networks[1] such that one can
make the system open, and resistant to abuse.  Obviously, it is not
possible to make any system based on trust completely free of abuse
(i.e. your best friend of 20 years might stab you in the back; no
system will see that coming), but it is possible to design a system in
that the amount of good you do in building up trust is outweighs the
amount of harm you can do in betraying people (i.e. it is easier to
lose trust than it is to gain it).

Also, correct me if I am wrong, but razor has a global notion of
trust:  i.e. Alice is trusted by everyone the same amount, as opposed
to each person having an individual opinion of Alice.  I tend to
believe that the latter type of system is far more resistent to abuse.

In anycase, the intent of my email was not to nitpick the razor system,
but to figure out how it works (if possible without reverse
engineering).  It wasn't clear from the docs that the TeS algorithm was
closed source, but now I will consider that question answered.

Thanks,

- Rob
.

[1] Shameless plug for the systems I've been working on that is 
about to be published : http://www.cs.umd.edu/projects/nice